HAL Id: inserm-01579134
https://www.hal.inserm.fr/inserm-01579134v2
Submitted on 26 Feb 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Comparison of inhomogeneity distortion correction
methods in diffusion MRI of the spinal cord
Haykel Snoussi, Emmanuel Caruyer, Olivier Commowick, Elise Bannier, Anne
Kerbrat, Christian Barillot
To cite this version:
Haykel Snoussi, Emmanuel Caruyer, Olivier Commowick, Elise Bannier, Anne Kerbrat, et al..
Com-parison of inhomogeneity distortion correction methods in diffusion MRI of the spinal cord. ESMRMB
- 34th Annual Scientific Meeting European Society for Magnetic Resonance in Medecine and Biology,
Oct 2017, Barcelone, Spain. �inserm-01579134v2�
Comparison of inhomogeneity distortion correction methods in
diffusion MRI of the spinal cord
Haykel Snoussi1, Emmanuel Caruyer1,2, Olivier Commowick1, Elise Bannier1,3, Anne Kerbrat3, Christian Barillot1,2
1
VisAGeS Reseach Unit, Inria Rennes - Bretagne Atlantique, France
2 CNRS, IRISA UMR 6074, Rennes, France 3
Centre Hospitalier Universitaire de Rennes, Neurinfo platform, France
1
Introduction
Diffusion MRI (dMRI) is a modality that describes the geometry of neural architecture. Diffusion images suffer from various artifacts originating from subject and physiological motion, eddy currents and B0-field inhomogeneity. These can severely affect image quality particularly in the spine region. However, strategies exist to correct these distortions, including co-registration, point spread function, phase field map and reversed gradient polarity method (RGPM). In this work, we focus on the comparative evaluation of correction methods using RGPM which provides best results [1]. More precisely, we compared Voss [1, 2] plus two other recent methods: Topup (FSL) [3] and HySCO (ACID/SPM) [4].
2
Subjects and Methods
Data acquisition: 116 dMRI datasets were acquired on 61 multiple sclerosis subjects and 9 healthy volunteers (between 1 and 3 yearly visits/subject) from Rennes and Montpellier hospitals (3T Siemens Verio and Skyra, respectively) using the following parameters: sagittal orientation, TR/TE=3600/90ms, 2x2x2mm3resolution,
6 b=0 and 30 non-collinear DWI (900 s.mm−2) plus one b=0 image with reverse phase-encoding direction. After a conservative quality control (see below), we kept 69 acquisitions.
Quality control : In order not to bias the distortion correction comparison, we eliminated images with too many artifacts not related to B0-field inhomogeneity (motion, ghosting) and acquisitions for which rigid reg-istration between anatomical and diffusion failed.
Comparing distortion correction methods: The whole spinal cord was segmented using the Spinal Cord Toolbox [5] on a T2-weighted scan to obtain a binary mask which is registered to the first b=0 volume. Cross-correlation was computed (Fig. 1) between each corrected dMRI and the rigidly registered T2 image (Fig. 2), only within the spinal cord region.
3
Results
ANOVA was performed on the cross-correlation scores, and confirmed that there is a significant improve-ment using correction distortion (F=19.8, p=1.2710−11). Further, a Tukeys test showed that both HySCO and Topup perform better than Voss and uncorrected (p<210−4). However there is no significant difference between Voss and no correction. This last result is in contradiction with [1] which included 3 datasets, while we report results on 69 acquisitions.
Another finding (data not reported) is that changing the order in which RGPM images are given as in-put (first F>H then H>F, or reverse) yields different results, albeit no significant difference was noted.
Fig. 2. Example of a corrected b=0 volume; the mask of the spine computed using T2 and registered to the diffusion image is overlaid
4
Discussion and Conclusion
This comparative study conducted on 69 images confirms that performing a distortion correction step gives better alignment with anatomical image which is important for the accuracy of the subsequent analysis; out of the 3 methods tested, HySCO and Topup provide best results.
Acknowledgments
MRI data acquisition was supported by the Neurinfo MRI research facility from the University of Rennes I. Neurinfo is granted by the the European Union (FEDER), the French State, the Brittany Council, Rennes Metropole, Inria, Inserm and the University Hospital of Rennes.
References
1. J. Cohen-Adad, H. Lundell, and S. Rossignol, “Distortion correction in spinal cord dti: what’s the best approach,” in Proceedings of the 17th Annual Meeting of ISMRM, Honolulu, USA, vol. 3178, 2009.
2. H. U. Voss, R. Watts, A. M. Ulu˘g, and D. Ballon, “Fiber tracking in the cervical spine and inferior brain regions with reversed gradient diffusion tensor imaging,” Magnetic resonance imaging, vol. 24, no. 3, pp. 231–239, 2006. 3. J. L. Andersson, S. Skare, and J. Ashburner, “How to correct susceptibility distortions in spin-echo echo-planar
images: application to diffusion tensor imaging,” Neuroimage, vol. 20, no. 2, pp. 870–888, 2003.
4. L. Ruthotto, H. Kugel, J. Olesch, B. Fischer, J. Modersitzki, M. Burger, and C. Wolters, “Diffeomorphic suscepti-bility artifact correction of diffusion-weighted magnetic resonance images,” Physics in Medicine & Biology, vol. 57, no. 18, p. 5715, 2012.
5. B. De Leener, S. L´evy, S. M. Dupont, V. S. Fonov, N. Stikov, D. L. Collins, V. Callot, and J. Cohen-Adad, “Sct: Spinal cord toolbox, an open-source software for processing spinal cord mri data,” Neuroimage, vol. 145, pp. 24–43, 2017.
